Randomized controlled trials of psychological treatments for eating disorders in the DSM-5 era: a systematic scoping review and evidence map

Abstract Background Existing reviews of psychological treatments for eating disorders have primarily addressed treatment efficacy, while the design and reporting characteristics of the underlying trials have not been systematically characterized. This scoping review aimed to address this gap in order to inform future trial design. Methods PubMed, PsycINFO, Embase, and CENTRAL were searched from January 1, 2013, to April 28, 2026. RCTs evaluating psychological interventions for eating disorders diagnosed according to DSM-5 or DSM-5-TR criteria were included. Results Eighty-one RCTs were included, and annual publication averages increased across periods. Most trials were conducted in Western countries and predominantly enrolled adult women. More than half included fewer than 100 participants, 30 were pilot or feasibility trials, 69 had follow-up periods of 12 months or less, and treatment-end assessment dropout rates frequently exceeded 20 ~ 30%. Binge-eating disorder was represented in 48 trials. Avoidant/restrictive food intake disorder appeared in one. No eligible trials addressed pica or rumination disorder. Unspecified feeding or eating disorder occurred within mixed samples. Cognitive-behavioral approaches were evaluated in 45 trials. Digital interventions were tested mainly against waitlist controls. Outcomes focused largely on symptoms, whereas caregiver impact and costs or resource use were rarely assessed. Fifty-five trials reported a qualifying sample-size rationale. Allocation concealment, treatment fidelity and harms remained incompletely reported. Conclusions The RCT literature on psychological treatments for eating disorders is expanding but remains largely exploratory. Future trials could prioritize broader populations and diagnoses, active comparator designs, outcomes valued by patients and service providers, and prospective registration with complete reporting.